Reninforment Learning with Bipedal Robot

Bipedal envs in Pybullet

BipedalBulletEnv
|
└───BipedalBaseEnv: implementaton of step, reset, reward function
│   └───BulletBaseEnv: implementaton of render function
│  
└───BipedalRobot: implementation of 
    └───URDFBaseRobot: implementation of loading urdf, applying torques, reset simulation in Pybullet

RL controller

Outputs joint torques generated by reinforcement learning.

conda create --name <env> python=3.7
conda activate <env>
pip install -r requirements.txt
python bipedal_rlctrl.py

EA controller

Outputs torques generated by pre-trained evolution algorithm.

conda activate <env>
python bipdal_eactrl.py

Debug URDF

conda activate <env>
python bipdal_debug.py